 So, first I'd like to thank the organizers for giving me the opportunity to speak with everyone today. So, I'm here from the Translational Genomics Research Institute, which is kind of one of these unintended spin-outs of NHGRI, Jeffrey Trent left NHGRI about 2002, and that's when they started designing the building and bringing together scientists focused on the concept which brings us here today. So, in essence, it's the goal of the institute, the goal of the research. And I'm going to describe some of the work that we're doing and really represent a few other people here, Dan von Hoff, who's our physician-in-chief and leads us on the clinical side. Myself and John Carpton lead the sequencing and informatics where the latter is more of my focus. So I'm going to talk mostly about the kind of applied clinical genomic center that we're working on and have been for some time, and it's going to follow up on some of the common sideric made with a focus on, misspelled, their oncology. And really it's focused in on kind of a premise of a question and a research question of Dan von Hoff's, is molecular profiling using these genomic technologies to increase availability of options for treating cancer patients? Do those perform better than standard of care? And the overall approach that we're using over the past two years is integrative analysis of whole transcriptome and whole genome sequence for treatment when no other clear options are available. And like the St. Jude talk yesterday, all of the sequencing and these projects are done in the context of clinical research protocols. And so this is something that's building on earlier work, and I put down there at the bottom this JCO article looking at molecular profiling through expression aphiarrays and their effectiveness versus kind of a standard of care using time to progression among other metrics to evaluate their effectiveness. But this is one of the things that, you know, as I see as collaborative opportunities I'm going to hit that. I am seeing a large number of genomes that aren't funded by NHGRI appear when individuals have late stage metastatic disease and there isn't a clear options and they're going to complete genomes, they're going to Illumina, they're going to various sequencing centers, having their tumor sequenced, a lot of times they're having the germline sequence, sometimes the RNA is sequenced and they're identifying targets through analysis, they're being placed on medicine after cleavalidation of any of those targets and many times there is an outcome measure that would be a time to progression, a survival. And what you're seeing here is this is occurring at a number of sites. And it's kind of this anecdotal network of individuals who try to work together and share common experiences. The science article yesterday was a science translational medicine article is a good example of that. And so it is one of those opportunities to get together, unique resources speaking towards having a data set and the problem is just getting them together but they actually have outcome as well when they're placed on a treatment and that's unique and a lot of cases they aren't funded by NHGRI. So we have a number of studies underway and so we've probably gone through about 50 patients in the past year and I'll walk through that process through a variety of studies, there's a series of studies that are forthcoming but yet to be announced and this is one of going to be kind of our core premises going forward. So I'm going to walk through this path and this is probably one of the more relevant slides and I'll leave the rest to examples which kind of shows how we're doing this working with our clinical partners and so there'll be a study with entry criteria and there's a surgical resection and this is one of the parts where there's involvement all along and like Eric said having frozen tumor is one of those things and having the right type and right amount of materials is important but at a first support layer for this that goes to a CLIA lab and currently we used Keras, GeneDX but we're also building one with a new recruitment to TGEN and essentially that's where the clinical path occurs and the sample is effectively kind of split along two paths with one path remaining at the CLIA center and then it going down to a second support layer which is a sequencing of a whole genome and RNA and that's one of the things we're looking at oncology and some of the objectives of the research protocols we're doing there are a lot of pharmaceutical developments that occur in the context of over expression right and so not always is a marker going to be DNA based. It may be DNA, RNA and protein they're all markers there along the same vein I'll show examples of that but the RNA helps support and steam along the analysis. There is quite a bit of work and it's a lot of time I spent on kind of integrative analysis of all these data sets and this feeds into informatics and for us we have historical databases that we have and that we've worked with others to develop that help in this kind of feeding into a board meeting of experts which then feed back to the first layer of support which goes into the question really what are you going to validate in a CLIA setting and to eventually go to the partners and kind of working on the treatment decision and then what's very important with this is get to an outcome because these are research studies and that's an important aspect but they also inform us and there is something that can be helpful in future uses in a way to kind of bring these all together. So in essence there are several layers of support here and you know a lot of the time does get spent actually on the connection points which is what we want to eliminate because these are individuals who will often have recurrent metastatic disease they've failed or various treatments they're probably on a treatment and this is all occurring anticipating that there may be need for a second therapy option or a third therapy option and so in essence we're trying to get all this done within a six to ten week twelve week timeline because that that's a point where you know obviously if the current treatment fails and it was in a research setting then there the next one it's going to be a bigger question so if you sequence the genome what targets can you identify and what can you do in these cases so that's kind of the end goal then the first two layers which you know really to identify those events and these are just a series of tumors that would kind of graphically represent as as as these circle diagrams but there's a you know a couple of key messages when you're looking at germline events there's been a lot of time looking at them a lot of times we're asking questions about a single event and there was a lot of discussion yesterday about black or white right and so what we see in a lot of the tumor genomes is that the variant specific to the tumor has a very high risk of being predicted prognostically or maybe even towards treatment and they're quite frequent a K-RAS mutation a B-RAS mutation these are things which are already in clinical settings they do have a lot but but here's a key they they will occur often together we're seeing emergence of these panels you know like 46 genes 60 genes 70 genes and you will not just always get one thing and the choice be obvious in that case because what I what we do see when we interact on the the the clinical site is somebody will ask a question well I have a PIK3CA mutation I have also a K-RAS mutation right so the question we were talking about in the breakout sessions in a way these are gene-gene interactions that do reach the the you know fairly far downstream into the clinical setting and there are questions that are hard to answer without additional research and so you know the idea here that it's just one and it'll be black or white it will likely require integrative analysis and at the end of the day you start to wonder whether or not there are you know needs for experts who become like radiologists or something that interpret series of events together right and understand the molecular biology those are things we talk about as down the road so what I'll do here is just walk through a couple of examples because I think the examples help bring the point and I think there's about 50 or so of these and and these are selected just to highlight some of the challenges and issues so first one is going to be a metastatic uterine transitional cell carcinoma this is part of a NFCR initial foundation cancer research project which Glenn Weiss is a the physician lead on this one and this has been sequenced on an alumina high seek and normal tumor we also did RNA seek and so when you're looking at things like panels and what would appear on panels you could select out various things but but had we done just a panel on this we would get a TP 53 mutation not surprisingly now when we go back and take a bigger look at all the richness of data that comes with whole genome sequencing we can start to look at different levels and in this graph I'm showing a lot of copy number amplications and deletions ascertain through the sequence and so you start to see more of these come up that you would recognize EGFR is amplified and and that's something that has to some in a research setting clinical utility there's mick which which confounds certain things there's pik3ca is an application but we also have TP 53 mutation and so it's considering these in the context together because there's a lot of research data out there speaking towards these all being important and it brings into the question of that decision-making process and so if we back away from this and we say let's take a look at some of these major events and EGFR FR application has a lot of mean towards predicting him you know a targeted therapy TP 53 is relevant and so it's what comes along with this but one of the things that don't mention is is a lot of the physicians we're working with are starting to make and think about what's not there and so in this case you know that this is one that I find difficult to answer a lot of the informatics say there isn't evidence for a K-RAS mutation and there are papers you know discussing the relevance or non-relevance and when we look at what comes out of somatic genomes they usually don't tell you what's not there right and a lot of times we don't because we don't always really have the right frameworks for describing that but there are decisions being you know considered in the context of that and so in this case a lot of the work that went on this was based on previous literature where you do have research on tumors of this type that are maybe negative for P-16 so there was a P-16 deletion there's P-53 EGFR and so this is one of those events that actually becomes more on the straightforward side right and so EGFR overexpression and application was actually clear validated because that's that that was one of the things that was believed to be the most helpful for the physician team and then the patient went on to text them up and based on the application and the absence of K-RAS mutation but the messages from this one are you know you do find these events which are a combination of a lot of different things together and I do start to think that with these targets target sequencing you're going to start to see combination events occur and people making decisions on combination of events and so you know the question is you know how do we kind of put together this information and even just starting with the somatic genome sharing or itself or the interpretive events so those aren't necessarily even descriptive but you know what's the outcome of these patients and so that's the second example I will give and then wrap this up so this is a collaboration with Life Technologies funded by their foundation US oncology, TGEN and CARIS diagnostics and basically focusing on triple negative breast cancer chemo resistant metastatic disease and so you know I'm going to show some of these graphs and they are sequence database but this is a normal tumor pair what we see is deletion of P10 which in a research setting helps inform about later downstream pathways. Now when I mentioned there's clear validation we don't really always or even that often focus on validating what we see exactly from the next gen a lot of times we're we're validating downstream for a P10 deletion where we may look at the expression but the end of day you know this one was actually going to have some clear validated results at downstream of the protein level and so one of the reasons we're needing to look at clear for informing some of this is that that that that it's not just for us going to be you know a sequence based assay because it does speak to broader context so in a combination of single risk factors that's another thing but you have these concepts it's so you know also one of these patients which had the P10 under expression we observed a BRAF what looked initially like an amplification but with the next gen sequencing we're able to reconstruct it and to double minute and see that it was amplified several fold I don't know the exact fold and so you see also some fish on this and so this individual here has a P10 deletion they have a BRAF amplification and these speak towards broader concepts which are really relevant in the development of some drugs and you know there there are directions you might go with each and so one of the things that comes out of these is a question combinations and so combinations are something that are often the realm of phase one trials and this particular patient was enrolled in the the Start Center clinical trial looking at PI3K, AKT, MTOR, and RASMAC pathways and in essence it's two therapies right because there are multiple events going on and this is a patient that you know the the the the clear results suggest might benefit by in this study these are actually some of the scans of the individual and one of the things that was notable is the the the the the start trial was you know observed a notable very remarkable as they described it regression based on that combination one of them the more notable ones of all the ones that they did and it had an impact in a number of ways including you know additional trials looking at this because this was the only triple negative breast cancer patient that was described and done within the context of this and this is presented at ASCO earlier so just some messages whole genome sequencing and cancer is being done I I think I'm hearing it more often in the context of treatment within open clinical protocols I think the yesterday there was indication of that you're seeing it often when you contact Illuminous Cleo labs I think they told me that 40% of the time it's going to be cancer for them the completes told me the same and then you get these hard drives moving around and and then you get these decisions being made in research settings and there are outcomes and so I think a major question is how do you put this together because this is really new territory it's it's something that does belong in research but does have broader scale impact later downstream and so there are few ways to to facilitate that and so those are kind of the the the key messages and there's a lot of people here I kind of highlighted them throughout and there's a lot of different sources and I couldn't fit them all in one slide so I I risk offending everyone over anyone specific but thank you all right thanks very much we have time for a couple of questions if there are any so thank you there's certainly a lot of enthusiasm in the oncology community for diagnostic tools to help make decisions and many people are doing it but I'm not aware of any rigorous studies have actually tried to demonstrate that the decisions that are being made are the correct ones are you aware of any such studies and what are TGEN's plans for demonstrating that these technologies are actually useful as opposed to just anecdotal so the the one that we point to for on the rigorous side would be the JCO paper with Dan von Hoff and where that became the the premise for target now now that big thing that the editorial that accompanied that pointed out correctly is this isn't a randomized trial and so Dan has said one thing before he will retire that he has to do and to make this credible and to really go forward is there is a need for a randomized trial going forward and I would say that it is probably the most important thing we can do is to really do that in a proper way and I would say going forward that is needed to get away from this anecdotal part because when you show anecdotal results and I don't like to they give you a wrong impression sometimes a lot of the time so I think that these are research protocols but more importantly there is a need for some sort of randomized trial on this and I think that's that's the big thing that I believe going forward will be the the initiative and the funding is being brought together for that in a larger scale yeah you correctly point out that the editorial was actually fairly critical and was written by Jim Doreshow deputy director of national cancer institute and yet company Keras has been formed and is actually selling this technology on the basis of this one non-randomized pilot study that's been criticized and I think it's just important as TGen moves forward in this direction that the studies really be rigorous so we can understand the utility of these of these diagnostics yeah and I think that that that critique really is at home with Dan because whenever he presents the paper and I didn't do it he also presents the editorial side by side and the the impact that that has it so that the need for the rigorous trial and Keras you know and what happened after that definitely serves as important lessons going forward but but I guess the the to emphasize that the fear is this is coming so there is a high importance to putting that on and I would say that that's that's the thing that we do talk about the most is is putting this together in the rigorous of really randomized trial and not having it in a way where variables or questions can change right so I know that you have chosen not to do the sequencing in a in a clinical lab but to to use clear confirmation of variance I'm just wondering from a practical point of view obviously you've got to confirm all of the negative findings all of the lack of amplification all of the lack of mutations if you're using this for a clinical decision so do you end up using a like a 49 gene panel something to confirm your negative results or how do you clinically confirm the negative results on the sequencing and how does that compare cost wise with just doing the initial sequencing in the clinic so that's a really good question so first thing I'll say is is it's nearly in my mind and I'm an informaticians on the statistical side I don't think it's possible to prove a negative but and so you know obviously people do act upon that but that's something that that is kind of hard to do I have this other belief which is the most interesting thing from whole genome sequencing has the highest error rate and we see that from from things like 1000 genomes loss of function there's a lot of logic to why that is and so the idea of how you interpret something when you have four million variants and millions of hypotheses asked I we do actually do much for sequencing in the alumina cleolab but we still validate it additionally because you know if you said where are the loss of function variants in that alumina analysis you're going to find a higher error rate of false positives in would you grab a random snip out of anywhere so the mentality for the that I have is is that it is helpful but even when we run our we we ship it to illuminants done in a cleolab we don't interpret the variants in the way and so for us now in terms of could we just run another panel we find a lot of events in our cases are actually things that might be missed by various panels an amplification might be missed by a 46 gene panel which goes to PCR exhaustion that's only really validated on mutations and indels right translocations do not appear effectively in most panels so in that context the panel is helpful for validation but you don't know which panel different panels will tell you different things and sometimes you do need to go further downstream I do have a lot of nervousness about going forward with results even you know when they look clear as day there may be something surprising so additional clear validation it is something that we're really thinking of and for us there will be the the exomes figured out and how you interpret variants and there's a lot of great people doing that but but we've kind of taken this path and we'll follow the lead of others for whole genome exome and direct interpretation because that's a little bit beyond this what we can do so I agree with you that the whole genome sequencing approach is much better than the limited panels but in the experience that I have the vast majority of tumors are highly contaminated with normal stroma and so how what percent of tumors do you think can actually be subjected to this type of analysis so a lot of my energy you know on and so forth we've been thinking about design so internally we will exome sequence plus some sort of medium pass whole genome sequencing plus RNA seeks so that we try to put depth where depth is needed we focus on insert sides outside of you know for the whole genome side because if we're going to interpret something in the middle of nowhere it might be a translocation or structural event and therefore sequence depth is an important as physical depth right so we try to basically design the whole genome assay around addressing that depth problem so that in coding regions we're 200 x 3 depth in intronic regions we might be low sequence depth but high physical coverage depth we're using combination of long of long insert and exomes and RNA seek because I find that that helps us get a little bit lower and the allelic frequency for the somatic variation with those 40 percent 50 percent tumor and then we the the research protocol specify you know really specific we take the the tumors based on whether or not we think we can be successful so that's where I go back to that surgical resection is actually essential and it does dictate what you can do later all right I think we need to move on thank you very much next speaker is marin schooner who's from the